Breakdown of broken-symmetry approach to exchange interaction

TL;DR

This paper critically examines the limitations of broken-symmetry approaches in calculating exchange interactions, revealing intrinsic issues related to covalency and proposing a multiconfigurational extension to improve accuracy.

Contribution

It identifies fundamental problems in the broken-symmetry method for low-spin states and introduces a multiconfigurational extension as a potential solution.

Findings

Error scales with covalency degree

Intrinsic issues due to single-determinant constraints

Proposed multiconfigurational extension as improvement

Abstract

Broken-symmetry (BS) approaches are widely employed to evaluate Heisenberg exchange parameters, primarily in combination with DFT calculations. For many magnetic materials, BS-DFT calculations give reasonable estimations of exchange parameters, although systematic failures have also been reported. While the latter were attributed to deficiencies of approximate exchange-correlation functional, we prove here by treating a simple model system that the broken-symmetry methodology has serious problems. Detailed analysis clarifies the intrinsic issue with the broken-symmetry treatment of low-spin states. It shows, in particular, that the error in the BS calculation of exchange parameter scales with the degree of covalency between the magnetic and the bridging orbitals. This is due to the constraint on the form of multiconfigurational state imposed by the BS determinant, a feature common to other single-reference methods too. As a possible tool to overcome this intrinsic drawback of single-determinant BS approaches, we propose their extension to a minimal multiconfigurational version.